WordGPT

Margin of error for n = 3,000 at 90% confidence

With a sample of 3,000 the margin of error at 90% confidence is about ±1.50 percentage points (worst case, p = 0.5).

How to work it out

  1. 1

    The margin of error for a proportion is z* times the standard error.

    E=z∗p(1−p)nE = z^{*}\sqrt{\frac{p(1-p)}{n}}
  2. 2

    For 90% confidence z* = 1.645. Use p = 0.5 for the worst case.

    E=1.6450.5×0.53000=0.01502E = 1.645\sqrt{\frac{0.5 \times 0.5}{3000}} = 0.01502
  3. 3

    As a percentage:

    E=±1.50%E = \pm 1.50\%
  4. 4

    Quadrupling the sample size only halves the margin of error, because n is under a square root.

LaTeX source of the formulas on this page
\[ E = z^{*}\sqrt{\frac{p(1-p)}{n}} = 1.50\% \]
\[ E = z^{*}\sqrt{\frac{p(1-p)}{n}} \]
\[ E = 1.645\sqrt{\frac{0.5 \times 0.5}{3000}} = 0.01502 \]
\[ E = \pm 1.50\% \]

Margin of error for n = 3,000 by proportion

Proportion pMargin of error
0.5±1.50%
0.4±1.47%
0.3±1.38%
0.2±1.20%
0.1±0.90%
0.05±0.65%

Margin of error for n = 3,000 at other confidence levels

ConfidenceMargin of error (p = 0.5)
80%±1.17%
90%±1.50%
95%±1.79%
98%±2.12%
99%±2.35%

Questions

What is the margin of error for 3,000 respondents?

About ±1.50 percentage points at 90% confidence.

How does the margin of error change with sample size?

It is proportional to 1/√n: a sample four times as large has half the margin of error.

What does a margin of error of ±1.5% mean?

If the poll were repeated many times, 90% of the intervals p̂ ± 1.5% would contain the true proportion.

Other sample sizes

Other confidence levels

    Margin of error for a sample of 3,000 at 90% confidence: ±1.50% | WordGPT